[boost] endless loop in dev_poll_reactor
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
boost (Ubuntu) |
Triaged
|
Medium
|
Unassigned |
Bug Description
Could you please apply the patch from:
https:/
Although the bug report suggested it was solaris specific, it can be reproduced on both Karmic and Lucid (I tried boost1.38 only, not boost1.40 but the issue should affect both).
The strace output is like this:
strace produces constantly output similar to this:
- - -
gettimeofday(
poll([{fd=3, events=POLLIN}, {fd=11, events=
events=
events=
{fd=18, events=POLLIN}, {fd=19, events=POLLIN}, {fd=7, events=POLLIN},
{fd=40, events=POLLIN}, {fd=8, events=POLLIN}], 12, 0) = 0 (Timeout)
read(7, 0x92813a0, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
read(7, 0x92813a0, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
read(16, 0x92b5470, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
gettimeofday(
poll([{fd=3, events=POLLIN}, {fd=11, events=
events=
events=
{fd=18, events=POLLIN}, {fd=19, events=POLLIN}, {fd=7, events=POLLIN},
{fd=40, events=POLLIN}, {fd=8, events=POLLIN}], 12, 0) = 0 (Timeout)
read(7, 0x92813a0, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
clock_gettime(
read(7, 0x92813a0, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
read(16, 0x92b5470, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
gettimeofday(
poll([{fd=3, events=POLLIN}, {fd=11, events=
events=
events=
{fd=18, events=POLLIN}, {fd=19, events=POLLIN}, {fd=7, events=POLLIN},
{fd=40, events=POLLIN}, {fd=8, events=POLLIN}], 12, 0) = 0 (Timeout)
read(7, 0x92813a0, 4096) = -1 EAGAIN (Resource temporarily
unavailable)
- - - (and many more, never ends)
Patch was applied on boost SVN here: /svn.boost. org/trac/ boost/changeset /58762
https:/